/log
3/9/2005
In my spare time I made a onscreen emulator/parser for my midi footpedal. The reason for this piece of equipment is still under development!
I also started messing with Jitter that could randomly crossfade through 20 different videofiles.
.:: Video Mixer
2/28/2005
Added the Final Project Page
2/23/2005
Interim Progress Report on Final Project:
I have been investigating Max more, as well as reading the Masataka Goto papers, and reviewing my old neural network source code. The first little Max snipplet is just something you can tap your mouse to get the bpm, and have the metronome play at that bpm. I am currently trying to tie that with live sound input, as well as introducing the hidden layer in the neural network with the activation and bias for each neural 'cell' in the hidden layer.
2/18/2005
Major revamp of the Sound Spatialization max file.
2/14/2005
First Max/MSP Project : Sound Spatialization
I have tried implementing a doppler effect simulator by just using the natural properties of the speed at which the sound travels. However, I put my own twist into it by simulating listening in stereo with delay and intensity difference, as well as a GUI to configure.
Some shortcomings: The intensity is still not calculated from the sound intensity decay (through air) curve, and the sound is choppy (probably due to lack of good interpolation), and the parameters are reset whenever it is loaded, so I need to devise a good preset system, but in the meantime, you can manually configure it before playing around with the motion input GUI.
Recommended settings:
- Sound Source: "2" unless you have a pretty consistent and long sound file to load and play
- Master Volume: start somewhere like at 0.5
- duration (of motion): try something like 3000 (milliseconds)
- head size (width and depth) : try 0.3 for a big 30cm head
- finally, you need to toggle the audio on to start playing.
.:: Max/MSP Spatialization Patch
EDIT: Update of the Spatialization Patch improves upon it by a large amount!
First off, default settings are autoloaded, and intensity is controlled through an inverse power relationship, which calculates the volume based on distance.
.:: Max/MSP Spatialization Patch version 2
2/7/2005
Some more random musings on my final project:
- Curtis Roads book (pg 520 - 531) for introduction
- Look into Leonardo magazine/journal
- Research Langston Library for Music and AI
- Dig up old Neural Network projects (from ICS 171)
- Learn about FFT
- Masataka Goto: Beat Tracking System
- Contact So Yamaoka for more ideas
- Contact various companies for their possible inputt
- Play around more with PortAudio/CoreAudio to increase familiarity with the system in preparation for the final implementation
2/4/2005
Since it seemed beneficial, I shrunk my project to a simple realtime delay, with much much better coding style and comments. I hope this will be helpful for somebody.
.:: padelay.c (much better comments!)
Furthermore, an even easier way to get started for MacOSX 10.3 people.
.:: Port Audio Delay XCode 1.5 Project
2/1/2005
Ideas for my research project:
Strongly influenced by the demo and discussion during the last class, I am thinking of making a realtime beat analyzer/guesser. This would have several portions, including spectral analysis, composition of rhythm from beats, and a neural network to help extrapolate the analyzed data to guess the next beat. If this can be done successfully, I should be able to at least be able to output the BPM, and with some assumptions, the meter as well. The program itself should be very complicated but boring. However, the possibilities are endless. By sending the midi signal for these parameters, it should be able to drive a drum machines and sequencers to another live rhythmical sound. This should make it possible to "jam" with your computer, and with a few more live input to control parameters in the way musicians communicate by gestures, it can be programmed to hold the beat. The other possibility is that the neural network will adapt itself to the beat, so not playing the constant pulses, as long as it's still at the right tempo, might allow the live performer to solo. Another interesting application would be to allow sidechain input for time-based effects such as delay, where the audio signal from the drum will automatically set the internal tempo of the delay. Finally, any other program, such as music visualizers can utilize the data generated by the realtime beat analyzer.
Research:
- Circular Logic InTime (link) - This is sort of what I want to do, but it only takes MIDI input.
1/18/2005
PROJECT 1: My goal for it was to have a working realtime effects processor for voice, guitar, etc. The first step was to create a delay. That was easy, but I actually went through and commented it along with simple UI. After that, I've added a tremolo effect and is in the process of adding a phaser. No GUI as of yet since I didn't have time to learn the MacOS X Cocoa framework.
.:: project1.c (commented code for an easy portaudio realtime CLI program template)
1/11/2005
Created website. This will probably be ugly on IE6 with the wonderfully bad support for PNG files. The site is currently barren, but I'm sure it will fill up with my projects in time. In the meantime, I have put up some links that may be of interest.
Anyways, my quest for finding fun and interesting uses for DSP starts!